The genus Sciarotricha gen. n. (Sciaridae) and the phylogeny of recent and fossil Sciaroidea (Diptera)

Transcription

1 The genus Sciarotricha gen. n. (Sciaridae) and the phylogeny of recent and fossil Sciaroidea (Diptera) HEIKKI HIPPA and PEKKA VILKAMAA Insect Syst.Evol. Hippa, H. and Vilkamaa, P.: The genus Sciarotricha gen.n. (Sciaridae) and the phylogeny of recent and fossil Sciaroidea (Diptera). Insect Syst. Evol. 36: Copenhagen, June, ISSN X. The phylogeny of the main groups of the Sciaroidea, including the fossil Antefungivoridae, Archizelmiridae, Mesosciophilidae, Pleciofungivoridae, Pleciomimidae, Protopleciidae and Bolitophilidae: Mangasinae, and an extant new taxon, was studied by parsimony analysis. Two cladistic analyses of seventy-eight morphological characters from adults were made. One analysis, with forty-one extant taxa in the ingroup and the other, with the addition of twelve fossil taxa, both produced two most parsimonious cladograms. The phylogenetic hypotheses obtained differed from each other, and in part also to a great extent from previous ones although most of the traditionally recognized groups appeared monophyletic, including the speciose Cecidomyiidae and Sciaridae. The Cecidomyiidae (fossil analysis) or the Keroplatidae-Ditomyiidae (extant analysis) appeared as the sister-group of the rest of the Sciaroidea. Following on from these analyses, we propose emending the current Sciaridae to include the following subfamilies: Archizelmirinae stat. n., Rangomaraminae stat. n., Sciarinae, Sciarosominae subfam. n. and Sciarotrichinae subfam. n. A new taxon from Namibia, Sciarotricha biloba gen. n., sp. n. is described, and, according to the phylogenetic analysis, is placed in the Sciaridae (Sciarotrichinae). The sister-group of the Sciaridae as newly defined is the Mycetophilidae group, in the extant analysis including the Mycetophilidae, Manotidae, Lygistorrhinidae, Pterogymnus and Sciaropota, and in the fossil analysis even including the Mesosciophilidae and the Ohakunea group (Ohakunea+ Colonomyia). Heikki Hippa, Swedish Museum of Natural History, PO Box 50007, S Stockholm, Sweden (heikki.hippa@nrm.se) Pekka Vilkamaa, Finnish Museum of Natural History, Zoological Museum, PO Box 17, FI University of Helsinki, Finland (pekka.vilkamaa@helsinki.fi) Introduction The Sciaroidea is a superfamily of the Diptera with a worldwide distribution, and includes about ten currently recognized extant families and six fossil families dating from the Lower Jurassic onwards (see Grimaldi & Blagoderov 2004). Some of the groups are extremely successful and speciose, such as the Sciaridae, Cecidomyiidae and Mycetophilidae, whilst others are speciespoor but nevertheless phylogenetically intriguing (the Heterotricha group, Rangomaramidae, Ohakunea, Colonomyia). There is a consensus regarding the placement of the Sciaroidea in the Bibionomorpha but not as regards the groups that should be included: for example, the Cecidomyiidae are sometimes excluded (Söli et al. 2000). In addition to the classic works of Hennig (1968, 1973) and Tuomikoski (1961, 1966), different hypotheses about the phylogeny of various groups of the Bibionomorpha or Sciaroidea have been published in recent decades, some using traditional methods of character analysis (Wood & Borkent 1989, Matile 1990, Blaschke-Berthold 1994, Shcherbakov et al. 1995, Jaschhof & Didham 2002, Grimaldi et al. 2003, Jaschhof 2004), others using cladistic computer methods (Amorim 1992, Oosterbroek & Courtney 1995, Matile 1997, Söli 1997). Blagoderov & Grimaldi (2004) presented a quantitative cladistic analysis of some subgroups of the Sciaroidea, but additionally gave a detailed discussion of various important characters previously used to infer phylogenies within the Sciar- Insect Systematics & Evolution (Group 6)

2 122 Hippa, H. & Vilkamaa, P. INSECT SYST. EVOL. 36:2 (2005) oidea. So far as our own work is concerned, the paper by Matile (1997) is most relevant as it included most of the groups that we analyse here and in a similar numerical analysis. There is little consensus between the various hypotheses regarding sciaroid phylogeny, perhaps the most striking disagreement being the question of whether or not the Cecidomyiidae and Sciaridae are sister-groups (Vilkamaa & Hippa 1998). The present study was prompted by recent advances in the field, including the publication of important papers by Chandler (2002) and Jaschhof (2004) on the enigmatic Heterotricha group of genera, by Jaschhof & Didham (2002) on the new family Rangomaramidae, by Grimaldi et al. (2003) on the Mesozoic Archizelmiridae, by Blagoderov & Grimaldi (2004) on the Sciaroidea in Cretaceous ambers, and, not least, by the discovery of an exceptional new taxon from Namibia, which is described below. Our aim was to obtain a phylogenetic hypothesis of the relationships between all the main groups of the Sciaroidea, including well-documented fossils, by means of a cladistic analysis using morphological characters from the adults, as the immature stages of many crucial taxa are unknown. Our main focus was on the Sciaridae and their phylogenetic position, but we also wanted to determine the phylogenetically correct placement for the problematic, unplaced genera within the Sciaroidea, including our new genus. Material and methods The specimens. - In the main, the material was studied with a compound microscope from slidemounted specimens, although pinned and ethanol/ glycerol-preserved material was used mainly when studying larger specimens and when different angles of view were necessary. Heterotricha in Baltic amber were studied with a stereomicroscope after cutting and polishing the blocks. Some of the material was available to us only from published descriptions in the literature. The material of Sciarotricha was collected with dry-collecting light traps and was subsequently stored in ethanol. Part of the material was studied in ethanol, where it is still stored, and part was mounted from ethanol on to microscope slides in euparal after dehydrating it in absolute ethanol. For morphological analysis, including the drawings, some specimens were treated with potassium hydroxide (KOH) before mounting in euparal. The structures of the male hypopygium were also studied from KOH-treated specimens under a stereomicroscope. The illustrations were made with the help of a drawing tube attached to a Leitz Diaplan compound microscope. The following abbreviations are used for institutes where material of the new species is located: BMNH The Natural History Museum, London, UK MZH Zoological Museum, Finnish Museum of Natural History, Helsinki, Finland NMNW National Museum of Namibia, Windhoek, Namibia NRM Swedish Museum of Natural History, Stockholm, Sweden Phylogenetic methods. - The data matrix (Table 2) was manipulated with the computer program WinClada, version 1.0 (Nixon, 1999). Phylogenetic relationships were studied by parsimony analysis using the computer program NONA, version 2.0 (Goloboff, 1993), used together with Win- Clada to search for the most parsimonious cladograms. The search parameters used with NONA were hold/100000, hold/100, mult*100 and TBR+max*. Characters were treated as unordered and with equal weights. With the above commands and settings, the program makes a heuristic search and swaps branches with tree bisectionreconnection. The unsupported nodes were collapsed to accept only unambiguous support in the strictest sense. The resulting cladograms and character optimizations were studied with WinClada. The ingroup (Table 1) of the first analysis included all the groups of recent Sciaroidea usually given a family or subfamily rank. Whenever possible, a species from the type-genus was chosen as a terminal, as well as species from all the available genera of obscure family placement (the Heterotricha group of Chandler 2002, Ohakunea, Colonomyia). Paleoplatyura was chosen as an additional Keroplatidae as it is considered to be very plesiomorphic (Chandler 2002). A few different structural types were chosen from the Sciaridae, 1) because the monophyly of the family is frequently questioned, 2) because some of them share apparent apomorphic character states with the new genus Sciarotricha as well as with some of the genera placed in the Heterotricha group, and 3) to help to resolve the postulated sister-group relation-

3 INSECT SYST. EVOL. 36:2 (2005) The genus Sciarotricha and phylogeny of Sciaroidea 123 Table 1. Terminals of the phylogenetic analyses For the fossil families, the nomenclature and generic affinities follow Evenhuis (1994). Outgroup: Bibionidae: Plecia sp. (Ecuador) Ingroup: Antefungivoridae : Antefungivora prima Rohdendorf (Mesozoic, Karatau); characters from Rohdendorf (1938). Archizelmiridae : Archimelzira americana Grimaldi et al. (Late Cretaceous, New Jersey amber); characters from Grimaldi et al. (2003). Archizelmiridae : Archizelmira baissa Grimaldi et al. (Early Cretaceous, Russia, Transbaikal); characters from Grimaldi et al. (2003). Archizelmiridae : Archizelmira kazachstanica Rohdendorf (Upper Jurassic, Kazakhstan, Karatau); characters from Grimaldi et al. (2003). Archizelmiridae : Burmazelmira aristica Grimaldi et al. (Mid-Cretaceous, Burmese amber); characters from Grimaldi et al. (2003). Archizelmiridae : Zelmiarcha lebanensis Grimaldi et al. (Cretaceous, Lebanese amber); characters from Grimaldi et al. (2003). Bolitophilidae: Bolitophila cinerea Meigen (Finland). Bolitophilidae, Mangasinae : Mangasinae gen. sp.; Characters from Kovalev (1986) and Chandler (2002). Cecidomyiidae, Catotrichinae: Catotricha americana (Felt) (North America). Cecidomyiidae, Catotrichinae: Catotricha fraterna Jaschhof (Australia). Cecidomyiidae, Lestremiinae: Lestremia cinerea Macquart (Sweden). Cecidomyiidae, Lestremiinae: Micromya sp. (Sweden). Cecidomyiidae, Porricondylinae: Porricondyla sp. (Finland). Cecidomyiidae, Cecidomyiinae: Cecidomyia sp. (Finland). Diadocidiidae: Diadocidia ferruginea Meigen (Finland). Ditomyiidae: Ditomyia fasciata (Meigen) (Finland). Keroplatidae: Keroplatus sesioides (Wahlberg) (Finland). Keroplatidae: Macrocera sp. (Finland). Keroplatidae: Arachnocampa luminosa (Skuse); characters from Matile (1990). Keroplatidae: Paleoplatyura sp. (Baltic amber; North America); characters from Meunier (1899), Johannsen (1910), Vockeroth (1981) and Chandler (2002). Lygistorrhinidae: Lygistorrhina sanctaecatharinae Thompson (North America). Mangasiidae : Mangas exilis Kovalev (Mesozoic, Russia, Transbaikal); characters from Kovalev (1986) and Chandler (2002). Manotidae: Promanota malaisei Tuomikoski (Burma). Mesosciophilidae : Mesosciophila venosa Rohdendorf (Upper Jurassic, Kazakhstan); characters from Rohdendorf (1946), and, for assistance with interpretation of the characters, Blagoderov (1993) was used. Mycetophilidae: Mycetophila fungorum (De Geer) (Finland). Mycetophilidae: Sciophila sp. (Finland). Pleciofungivoridae : Pleciofungivora latipennis Rohdendorf (Mesozoic, Karatau); characters from Rohdendorf (1938). Pleciomimidae : Pleciomima sepulta Rohdendorf (Mesozoic, Karatau); characters from Rohdendorf (1938). ship between the Sciaridae and the Cecidomyiidae (Wood & Borkent 1989, Oosterbroek & Courtney 1995) or between the Rangomaramidae + Cecidomyiidae and the Sciaridae (Jaschhof & Didham 2002). In the second analysis, the fossil families Antefungivoridae, Archizelmiridae, Mesosciophilidae, Pleciofungivoridae, Pleciomimidae, Protopleciidae and Bolitophilidae: Mangasinae were included in addition to the recent taxa. The Eoditomyiidae (Ansorge 1996) was not included because we suspect that it may belong to an older clade than our outgroup+ingroup. The genus Plecia (Bibionidae), which is commonly regarded as lying outside the Sciaroidea and in the Bibionomorpha (e.g. Amorim 1992), was chosen as an outgroup. The characters were taken only from adults because the immature stages are poorly known in many of the crucial groups studied. The male genitalia of Sciaroidea are extremely variable, and due to difficulties in interpreting homologous structures (see Jaschhof & Didham 2002, Jaschhof 2004) they are mostly omitted from the analysis. Characters used in the phylogenetic analysis. - With many of the fossils, essentially only the wing vein characters are known and available for cod-

4 124 Hippa, H. & Vilkamaa, P. INSECT SYST. EVOL. 36:2 (2005) Table 2. Data matrix for phylogentic analysis. A=0/1 polymorphism Plecia sp Archizelmira baissa ??????0?????? ?1? ???010???00?????-?????????? Archizelmira kazachtanica 1?0100??????0?????? ? ???010???00?????-?????????? Archimelzira americana ?0?0??0??0??? ?1? ???010???001????-?????????? Burmazelmira aristica ?0??0??0??? ?1? ???010??10010??0-?????????? Zelmiarcha lebanensis ?0?0??0??0??? ?1? ???010??0001??10-?????????? Sciara thomae Zygoneura sciarina Schwenckfeldina carbonaria Pnyxia sp Rangomarama edwardsi A Rangomarama humboldti A Rangomarama matilei A AA Colonomyia albicaulis ?01001 Ohakunea bicolor Sciaropota japonica ?00? Sciarotricha biloba A Sciarosoma borealis ???????? Nepaletricha sp ? ?-1-?????????? Kenyatricha elgon ? ??1??????????? Chiletricha marginata Insulatricha hippai Rhynchoheterotricha sp Anisotricha novaezealandiae ?0 Afrotricha relicta ?000?? ??0-?? Heterotricha takkae ? ??0-?? Pterogymnus elongatus ? Mesosciophila 1??00?00??00100?1??00110?0? ? ?????10???????????0?0 Promanota malaisei Mycetophila fungorum Sciophila sp Lygistorrhina s-catharinae ? Ditomyia fasciata Keroplatus sesioides Paleoplatyura sp ?1-100?1? ? Macrocera sp Arachnocampa luminosa ? ? Diadocidia ferruginea Bolitophila cinerea Cecidomyia sp Porricondyla sp Micromya sp ? Lestremia cinerea ? Catotricha americana 100? ? Catotricha fraterna 000? ? "Mangasinae sp."???0???????????????101?1?0?? ? ???10?????????????????????? Pleciomima sepulta???????????????????001???0??? ? ????00?????????????????????? Pleciofungivora latipennnis???????????????????001???0??? ?100001?100??????00?????????????????????? Antefungivora prima???????????????????001???0??? ? ????10?????????????????????? Archiplesiomima obtusipennis???????????????????01????1?? ? ?????00?????????????????????? Mangas exilis???????????????????00??1????? ?00000??1????????00?????????????????????? ing. We have adopted Chandler s (2002) interpretation of the wing venation, but use r-m instead of ta and m-cu instead of tb. 0. Male holoptic (0) (Fig. 3A); dichoptic (1) (Vockeroth 1981: Figs 4, 5, 7, 9). 1. Ocelli in an equilateral triangle (0) (Fig. 3A, B); in a transverse row, or the distance between posterior ocelli much greater than the distance between anterior and posterior ocelli (1) (Söli 1997: Fig.1A).

5 INSECT SYST. EVOL. 36:2 (2005) The genus Sciarotricha and phylogeny of Sciaroidea Antennal flagellum evenly broad or almost so (0) (Vockeroth 1981: Figs 3 7, 9, 10); flagellum strongly tapering, the last flagellomeres several times thinner than the basal ones (1) (Grimaldi et al. 2003: Figs 8:1, 3, 9: 1, 10:2, 3). 3. The node of the basal antennal flagellomere long, at least 5 times as long as broad (0) (Vockeroth 1981: Fig. 2); short, at most 4 times as long as broad (1) (Fig. 3A, Vockeroth 1981: Figs 1, 3-10). The difference between these states is usually clear. Rangomarama may be a borderline case. We have coded it as 1. Coding as 0 or? does not affect the trees. Catotricha is coded as? because of the difficulty in defining the node in a manner homologous with the other taxa with an elongate flagellomere. 4. Antennal flagellomeres single-noded (0) (Fig. 3A, 4B; Vockeroth 1981: Figs 3 7, 9, 10); double-noded (1) (Jaschhof 1998: Figs 13 e, 23 c; Gagné 1981: Figs 60 64). In many Cecidomyiidae, Cecidomyiinae, the flagellomeres have a double node, the two parts being separated by a constriction. In Cecidomyiidae, Catotrichinae, there is also a double node but the more apical one seems to be more of a setose swelling on the neck of the flagellomere. In many other Cecidomyiidae there is a more a less distinct non-setose swelling (see Jaschhof 1998) on the apical part of the neck of the flagellomere. We have coded all these cases as double-noded. 5. Antennal flagellomeres with monoporous sensillae (0) (Fig. 4A, B; Gagné 1981: Figs 39, 41, 42, 43, 44); with polyporous sensillae or circumfilia (1) (Gagné 1981: Figs 40, 46, 56, 60). These setae can be scattered, or they tend to appear in whorls in the Cecidomyiidae and some Bibionidae. No distinction between these cases has been made. We do not share the opinion of Jaschhof (2000) that the longer seta-like vestiture of the Cecidomyiidae, including his hooded sensillae, are not setae or trichoid sensillae. We have interpreted the structures called scales as Grimaldi et al. (2003) have done, Fig. 8:3, as strong setae. 6. Maxillary palp with 5 palpomeres (0); (Fig. 4C; Söli 1997: Fig. 8A, F); with 4 palpomeres (1) (Jaschhof 1998: Fig. 1); with 3 palpomeres or less (2) (Söli 1997: Fig. 8D, E). The complete or maximum number of palpomeres in the Sciaroidea is 5 (cf. Jaschhof 2000). The reduced number is the result of the obliteration of the basal segment by union with the original palpomere 2, or by total obliteration. Even this basal segment may be obscured because of its union with the original palpomere 3, which is apparently always present. The reduced number may even be the result of obliteration/union of the apical segment/s. 7. Palpomere 4 attached at the apex of palpomere 3 (0) (Fig. 4C); attached on the ventral side of palpomere 3 and far from its apex (1) (Vockeroth 1981: Fig. 8). 8. Head normal (0) (Fig 3; Vockeroth 1981: Figs 6, 7); long and deflexed (1) (Shaw 1948: Figs 16 21, Vockeroth 1981: Fig. 9). 9. Scutum with a non-setose stripe at least between lateral and dorsocentral-acrostichal setae (0) (Fig. 5A; Söli 1997: Fig. 13; Hutson et al. 1980: Figs 18-19); scutum completely setose (1) (Hutson et al. 1980: Figs 20-23). 10. Mesepimeron posteroventrally extending to posterior margin of katepisternum and ventral end of pleurotergite (0) (Fig. 5A, B, C); abbreviated and not reaching the ventral margin of pleura (1) (Shaw 1948: Fig. 9). 11. The anterior/anteroventral margin of mesepimeron ends ventrally opposite the middle coxa (0) (Fig. 5A); ends at the anterodorsal corner of episternum 3 (1) (Fig. 5B, C). This is a character which separates the Sciaridae in the traditional sense from other Sciaroidea, as already pointed out by Shaw (1948) and Matile (1990). 12. Laterotergite flat (0) (Fig. 5A); bulging (1) (Söli 1997: Figs 14A, C, D). 13. Anteroventral margin of laterotergite straight to slightly curved (0) (Shaw 1948: Figs I:2-7); roundly curved on entire anteroventral half (Fig. 5A, B, C; Shaw 1948: Fig I:1). 14. Suture between laterotergite and mediotergite distinct (0); absent (1). The absence of a clear suture between laterotergite and mediotergite was used as one of the characters of the Rangomaramidae by Jaschhof & Didham (2002). 15. Laterotergite non-setose (0) (Fig. 5B, C); setose (1) (Fig. 5A). 16. Pleural pit well-developed (0) (Fig. 5A, C); weakly developed or absent (1) (Fig. 5B).

6 126 Hippa, H. & Vilkamaa, P. INSECT SYST. EVOL. 36:2 (2005) The origin or invagination point of the pleural apodeme at the junction of the anepisternum, anepimeron and katepisternum may be open or closed. When open it is called the pleural pit. All intermediate stages occur within the Sciaroidea or Bibionomorpha. For present purposes, we classify the pleural pit as welldeveloped when it is large, exposed in lateral view, and when the anterior margin of the anepisternum is interrupted by the pit. In these cases the pleural apodeme is also a long, curved, funnel-like structure. 17. Episternum 3 non-setose (0) (Fig. 5A, B, C); setose (1). 18. Phragma developed, intruding into the base of the abdomen (0) (Figs 5A, B, C; Chandler 2002: Figs ); phragma undeveloped, not intruding into the abdomen (1) (Chandler 2002: Fig. 101). In most Sciaroidea the posteroventral part of the metanotum appears as a bulge of variable size and intrudes more or less deeply into the base of the abdomen (cf Chandler 2002). It is very prominent in the Cecidomyiidae and Sciaridae. In the present analysis, only Mycetophila, Sciophila, Lygistorrhina, Pterogymnus, Sciaropota and Promanota have the phragma undeveloped. 19. Pterostigma absent (0) (Fig. 5D); present (1) (Vockeroth 1981: Fig. 13). 20. Wing shape elongate (0) (Fig. 5D); round (1) (Grimaldi et al. (2003): Figs 1:1, 2, 4: 1, 2). The usual wing shape in the Sciaroidea is elongate, much longer than wide, but roundish forms occur here and there. In the present analysis the roundish form occurs only in the two species of Archizelmira (see Grimaldi et al. 2003). 21. Anal lobe of wing 90 degrees or more (0) (Fig. 5D); anal lobe of wing less than 90 degrees, i.e. its margin towards wing base recurrent (1) (Hardy 1981: Figs 6, 7, 9, 10, 11; Grimaldi et al. 2003: Figs 7:4. 8:4, 9:2, 10:1). The very strong anal lobe (1) occurs in the present analysis only in some Archizelmiridae and in the Bibionidae. It is also prominent in some Keroplatidae (Vockeroth 1981: Fig.16) but not in the same way as in the Bibionidae and Archizelmiridae. 22. Wing membrane setose (0) (Vockeroth 1981: Figs 11 14, 45).); non-setose (Fig. 5D). The wing membrane is coded as setose even if the setose area is very restricted, as in Mycetophila fungorum which we have studied. 23. The section of costa beyond the tip of R5 long, several times as long as its width (0) (Vockeroth 1981: Figs 13, 14, 15); short, at most three times as long as wide (1) (Fig. 3 D; Vockeroth 1981: Fig 11); absent or almost so (2) (Vockeroth 1981: Fig. 12). 24. Costal break absent (0) (Fig 5D; Gagné 1981: Fig. 23); present (1) (Gagné 1981: Figs 12, 14 22). In the Cecidomyiidae the costa continues as a faint vein-like thickening from the thick anterior portion ending at the tip of R5 or between R5 and M and around the posterior wing margin. The thickened posterior margin can even be seen in some other Sciaroidea. In most Cecidomyiidae there is a clear break between the thick anterior and thin posterior parts. In the Catotrichinae the break appears only as a break in the marginal setosity of the wing, and it is coded as for other Cecidomyiidae. We have not observed a costal break or a break in the setosity in any other sciaroid. 25. Subcosta ending in the costa (0) (Vockeroth 1981; Figs 13, 16, 17, 18, 19, 20); ending free (1) (Fig. 5D, Vockeroth 1981; Figs 11, 12, 15, 21, 22). 26. Subcosta setose (0) (Vockeroth 1981: Fig. 16); non-setose (1) (Fig. 5D). 27. Vein sc-r present (0) (Tuomikoski 1966: Fig. 5; Vockeroth 1981: Figs 13, 48); absent (1) (Fig. 5D). We consider Promanota, in which sc appears to end in R, to have sc-r. 28. Geniculus radialis absent (0) (Hardy 1981: Figs 6 11); present (1) (Fig. 5D, E). The base of R at the humeral cross vein is either straight or there is a step-like shift, the geniculus radialis, towards the costa at the humeral cross vein. 29. Vein R1 extending well into the apical half of the wing (0) (Vockeroth 1981: 11, 12, 13); extending to the middle of the wing (1) (Fig. 5D); extending only over the basal half of the wing (2) (Gagné 1981: Fig. 12). 30. R1 long (0) (Fig. 5A; Vockeroth 1981: Fig. 59); short (1) (Gagné 1981: Fig. 12). The length of R is very variable. It is only coded here as short in those cases where it is very short, transverse or transversely diagonal,

7 INSECT SYST. EVOL. 36:2 (2005) The genus Sciarotricha and phylogeny of Sciaroidea 127 and not longer than Rs. 31. Vein R4 present (0) (Vockeroth 1981: Figs 11 m- 13, 16 20, 23-26); absent (1) (Fig. 5A; Vockeroth 1981: Figs 14, 15, 22). 32. R4 ending in costa (0) (Vockeroth 1981: Figs 11, 12, 16, 17,19); in R1 (1) (Vockeroth 1981: Figs 13, 18, 23, 24). 33. R5 or R4+5 curved (0) (Fig. 5A); straight (1) (Vockeroth 1981: Fig. 47). R5 is usually curved, following the curvature of the costa, but in a few cases it is straight or almost so. 34. Vein Rs diagonal (0) (Vockeroth 1981: Figs 11 13, 15 20); transverse (1) (Vockeroth 1981: Figs 14, 38,); recurrent (2) (Grimaldi et al. 2003: Figs 4:1, 2, 3). The distinction between these states is usually clear. Rangomarama matilei is coded as diagonal but is a borderline case. The other species of Rangomarama have a distinctly transverse Rs. The distinction between states 1 and 2 may be obscure in some cases in the Sciaridae. A clearly recurrent Rs ( slanted in Grimaldi et al. 2003) seems to be present only in the Archizelmiridae which are thus the only cases that we have coded as state Vein Rs normal (0) (Fig. 5A; Vockeroth 1981: Figs 11 20); unusually long, 3 to 4 times as long as the width of costal cell (1) (Jaschhof & Didham 2002: Figs 5 7). State 1 is coded only for Rangomarama. In Diadocidia and some Keroplatidae, Rs is also long but it is very diagonal. The character state would perhaps be better described as an unusually long distance between R1 and R Position of vein Rs at the middle of the wing (0) (Vockeroth 1981: Figs 11 13); in the basal half of the wing (1) (Fig. 5A). These states are not always completely clear. The borderline cases are resolved by subjective estimates rather than by exact measurements. 37. Base of vein Rs (junction with R) normal, sclerotized (0) (Vockeroth 1981: Fig. 11); unsclerotized (1) (Vockeroth 1981: Figs 14, 16). The unsclerotized break at the base of Rs is characteristic of the Ditomyiidae and Diadocidiidae, and also of many Keroplatidae. 38. Vein r-m transverse (0) (Vockeroth 1981: Figs 12-14, 19); longitudinal (1) (Fig. 5A). We have coded r-m as transverse in cases when it is strictly transverse and when it is obliquely transverse. It is coded as longitudinal when r-m seems to be a continuation of R4+5 without any angle. It is coded as r-m absent when R and M are united or connected with a scarcely discernible r-m. Cases in which the vein is absent are coded as inapplicable ( - ). 39. Vein r-m normal, long (0) (Fig. 5D; Vockeroth 1981: Figs 12, 13, 14, 19); very short or absent (1) (Vockeroth 1981: Figs 11, 15 18, 20). An absent r-m means that a part of R and M are in direct contact at some point or are confluent over a shorter or longer distance. In the latter case there is a vein frm (character 40). 40. Vein frm absent (0) (Fig. 5A); present (1) (Vockeroth 1981: Figs 15 18, 20). When vein r-m is absent, there may be an unusual vein portion, frm, between stm and Rs. When R and M are touching only at one point, we have coded it as state Vein stm short (0) (Fig. 5A; Vockeroth 1981: Figs 31, 71; Chandler 2002: Figs 4, 11, 15); long (1) (Chandler 2002: Figs 52, 53, 63). Vein stm has been coded as short when it is distinctly shorter than the medial fork, and as long when it is of approximately the same length or distinctly longer. For a few borderline cases in the Heterotricha group, we follow Jaschhof s (2004) interpretation. Those cases in the Cecidomyiidae in which M is unforked are coded as 1 because we regard it as obvious that they are derived by an extreme lengthening of stm (cf. Jaschhof 1998: Fig. 43g) or by the obliteration of a short M1 (cf. Panelius 1965: Fig. 7a). 42. Vein portion M+ absent (0) (Fig, 5D; Vockeroth 1981: Figs 12, 38); present (1) (Vockeroth 1981: Figs 13, 19). M+ appears on M between veins r-m and bmcu when these are not contiguous. 43. Basal part of vein M present (0) (Vockeroth 1981: Fig. 19); present as a non-sclerotized fold (1) (Chandler 2002): Figs 37 38); absent (2) (Fig. 5D; Vockeroth 1981: Fig. 11) In most Sciaroidea the basal part of M is absent (Chandler 2002). In the Bibionidae it is a normally sclerotized vein. In some Sciaroidea it is clearly discernible as a weakly or partly weakly sclerotized vein which is more than simply a longitudinal fold on the

8 128 Hippa, H. & Vilkamaa, P. INSECT SYST. EVOL. 36:2 (2005) wing membrane. All these cases are coded as 0. The folds shown by Chandler (2002) in many species of the Heterotricha group, probably visible only in dry unmounted specimens, are coded as Vein m-cu connected to CuA1 (0) (Vockeroth 1981: Fig. 12); to stcu (1) (Fig. 5D; Vockeroth 1981: Fig. 71). 45. Vein m-cu transverse (0); oblique (1); parallel with R (2); converging with R towards wing base (3). 46. CuA1 and CuA2 divergent (0); parallel on basal two thirds (1). 47. stcu short, much shorter than CuA2 (0) (Fig. 5D); long, longer than CuA2 (Vockeroth 1981: Fig. 38) 48. CuA2 evenly curved (0) (Fig. 5D); sinuous (1) (Grimaldi et al. 2003: Figs 9:2, 10:1) This character was used by Grimaldi et al. (2003) as a synapomorphy of Archimelzira and Burmazelmira. 49. Vein A1 extending to wing margin (0) (Vockeroth 1981: Figs 11, 12); abbreviated (1) (Fig. 5D; Vockeroth 1981: Figs 15, 38). 50. Basal part of vein 1A normal, thin and poorly sclerotized (0) (Fig. 5D); thickened and sclerotized (1) (Jaschhof & Didham 2002: Figs 5 7). 51. Vein M dorsally setose (0); non-setose (1). 52. Vein CuA1 dorsally setose (0); non-setose (1). 53. Vein CuA2 dorsally setose (0); non-setose (1). A thickened and abruptly truncated basal portion of A1 is a characteristic of Rangomarama. There are faint indications of a similar development in Sciarosoma and Sciaropota. 54. Veins R4+5 and M1 parallel or divergent (0) (Fig.5D; Vockeroth 1981: Fig. 15); convergent towards wing margin (1) (Gagné 1981: Figs 18, 20). 55. Basal cell parallel-sided or widening towards apex (0); widening towards base (1). 56. Tibial spur/s present (0); absent (1). 57. Tibial organ absent (0); present (1) (Fig. 4D; Söli 1997: Fig. 20 A - E). 58. Tibiae medially with normal microtrichia (0); without normal microtrichia (1). The Cecidomyiidae differ from other Sciaroidea by having short, normal microtrichia along the whole tibia and lacking the enlarged trichia (non-socketed setae) in addition to setae (socketed setae). In other Sciaroidea, normal microtrichia are found only at the base of the tibia, and rarely also in other places among enlarged trichia. 59. The longer tibial vestiture with socketed setae only (0); with socketed setae and enlarged microtrichia which are as large as the setae (1) (Fig. 4E, D). No distinction has generally been made in the literature between different types of tibial vestiture except for the stronger spines and the weaker hairs or setae. The element usually called seta consists of two different kinds, those with sockets (trichoid sensillae) and those without sockets. The latter are probably enlarged microtrichia. 60. Tibial and tarsal setae normal (0) (Fig. 4D, E); scale-like (1) (Jaschhof 1998: Fig. 8D). The scale-like vestiture occurs only in Cecidomyiidae. The scales can occur on all body parts. When necessary, we have chosen the fore tibia as a reference point, or in extreme cases the fore basitarsomere. 61. Tibial vestiture normal (0) (Söli et al. 2000: Fig. 27); in rows (1) (Söli et al. 2000: Figs 26, 29-30). The arrangement of the tibial vestiture is a commonly-used character in sciaroid systematics. In most cases the character states are quite clear, but there are also cases which are difficult to judge. We have regarded a taxon as belonging to state 1 if there is a clear indication that the vestiture is arranged in rows at least on the apical part of the fore tibia. 62. Coxae short (0) (Jaschhof 1998: Fig. 3); long (1) (Fig.4A). We have not used any special measurements to estimate the states of this character. There is a clear difference between the Bibionidae and Cecidomyiidae, which have the coxae scarcely longer than broad, and the other Sciaroidea, which have them much longer that broad, especially the fore coxa. 63. Basitarsomeres normal, longer than tarsomere 2 (0) (Gagné 1981: Fig. 80); basitarsomeres short, shorter than tarsomere 2 (1) (Gagné 1981: Fig. 81). State 1 is found only in the Cecidomyiidae and is traditionally used to separate the Cecidomyiinae and Porricondylinae from the other groups. 64. Abdominal tergites with plaques (0) (Jaschhof 2000: Fig. 17); without plaques (1). The plaques appear as roundish or oval,

9 INSECT SYST. EVOL. 36:2 (2005) The genus Sciarotricha and phylogeny of Sciaroidea 129 sharply delimited, more or less distinctly depressed areas without the normal surface structure of the tergites. They can be concolorous with the adjacent area of the tergite, but usually they are paler, but sometimes even darker. They are usually easily seen, but in some cases are very difficult to discern, e.g. Sciaropota. 65. Abdominal tergites with two or more pairs of plaques (0) (Jaschhof 2000: Fig. 17); with one pair of plaques (1). 66. Posterior margin of male tergite 9 simple (0) (Chandler 2002: Fig. 18); with prominent ornamented lobes (1) (Chandler 2002: Figs 33, 50). 67. The strong setae/megasetae on male tergite 9 simple (0); double or cleft (1) (Jaschhof & Didham 2002: Fig. 30). The cleft enlarged setae on male tergite 9 were noted by Jaschhof & Didham (2002) in one species of Rangomarama. We found that there are similar setae in Chiletricha and Rhynchoheterotricha. They do not occur in Ohakunea in which there are enlarged setae in the same position. 68. Tegmen absent (0); present (1) (Steffan 1981: Figs 20 23; Gagné 1981: Figs 82 87; Jaschhof & Didham 2002: Fig. 71). By tegmen we mean a plate formed by medially fused parameres. A tegmen occurs in all the Cecidomyiidae and Sciaridae and in two species of the Rangomaramidae. This structure has not been studied in a large number of the groups included in the present analysis. 69. Aedeagal teeth absent (0) (Fig. 6A, C); present (1) (Vilkamaa 2000: Fig. 9B). In most genera of the Sciaridae, the apical part of the aedeagus bears small pointed tooth-like structures. These are clearly modified microtrichia and a complete morphological series from normal microtrichia to teeth can be seen in the family. The presence of these teeth was listed as a synapomorphy of Sciaridae by Blaschke-Berthold (1994) The microtrichia on the aedeagus can be found in many groups of the Sciaroidea, but they are often difficult to observe. They are present in at least most Cecidomyiidae, and in some cases modifications to the teeth can be seen (Jaschhof 1998: Fig. 43d and Fig. 74) though not in the ingroup taxa in the present analysis. In Rangomarama there are microtrichia in the same way as in many Sciaridae and at least as in Sciarosoma. 70. Female sternum 8 normal (0) (Fig. 7B); posteriorly high and concave (1) (Chandler 2002: Figs 25, 26, 35). 71. Female sternum 8 with normal setae (0) (Fig. 7B); with a dense fringe of posteriorly directed setae (1) (Chandler 2002: Figs 25, 26, 36). 72. Female sternum 9 normal (0) (Figs 7A, B); unusually large and sclerotized (1) (Chandler 2002: Figs 10, 25, 36). 73. Female tergite 10 with normal hairs (0) (Fig. 7B); with densely placed, long, golden hairs (1) (Chandler 2002: Figs 8, 10). 74. Female cercus two-segmented (0) (Fig. 7A, B); one-segmented or the apical segment greatly reduced in size (1) (Gagné 1981: Fig. 121). 75. Female cercal segment 1 normal (0) (Fig. 7B); produced apicodorsally (1) (Chandler 2002: Figs 25, 26). The basal segment of the cercus is umodified in all the taxa we studied, except in Chiletricha and Rhynchoheterotricha. In the former, the segment is bluntly produced apicodorsally, whilst in the latter there is a sclerotized curved horn-like projection. 76. Female cercal segment 2 simple (0) (Fig. 7A, B); modified (1) (Chandler 2002: Figs 8, 10). The cercal segment 2 is very uniformly elongate-oval in the Sciaroidea. In Colonomyia it is very narrow and finger-like apicolaterally on segment 1. In Heterotricha and Rhynchoheterotricha the apical part is curved laterad in the former, dorsolaterad in the latter. 77. Female spermathecae sclerotized (0) (Fig. 7A, B); unsclerotized (1). Results and discussion The analysis with only the extant taxa in the ingroup produced two most parsimonious cladograms, the strict consensus of which (241 steps, CI 36, RI 73) was fully resolved except for a trichotomy in one of the two main clades. However, there was much homoplasy in the data. The clade including the Keroplatidae and Ditomyiidae (node A, supported by five character state changes) appeared as the sister-group of all other ingroup taxa (Fig. 1). Our data supported a hypothesis of a close relationship between Keroplatidae and Ditomyiidae (B, three character state changes) but could not resolve the exact relationships among

10 130 Hippa, H. & Vilkamaa, P. INSECT SYST. EVOL. 36:2 (2005) Fig. 1. Phylogeny of the extant Sciaroidea. The strict consensus of two most parsimonious cladograms (241 steps, CI 36, RI 73) obtained with the program NONA. Letters above hatch marks refer to clades discussed in the text, numbers above hatch marks (open = homoplaseous, black = unique) refer to characters, numbers under hatch marks refer to state changes to the state indicated. Only unambiguous changes are shown.

11 INSECT SYST. EVOL. 36:2 (2005) The genus Sciarotricha and phylogeny of Sciaroidea 131 these. This result contrasted with the cladistic result of Matile (1997), based on 30 morphological characters, which suggested a sister-group relationship between the Keroplatidae and Diadocidiidae, whereas the Ditomyiidae had a more apical position in the phylogenetic system. The other main clade, (node C), including the rest of the ingroup terminals, showed Bolitophila and Diadocidia successively in the most basal positions (nodes D and E). The Heterotricha group of genera appeared polyphyletic, to an even greater extent than suggested by Chandler (2002), but none of its elements appeared as the sister-group of the Sciaridae, as suggested by Chandler (2002). Heterotricha itself and Anisotricha formed the sister-group of Colonomyia + Ohakunea (node F). The two latter genera, which have been without a clear family assignment, were sister-groups as had been conjectured by Jaschhof & Hippa (2002), Chandler (2002) and Hippa & Jaschhof (2004). Six of the genera in the Heterotricha group, including the new Insulatricha Jaschhof (2004), formed a monophyletic group which appeared as the sister-group of the Cecidomyiidae + (Mycetophilidae group of (sub)families + Sciaridae) in our new concept (node G). The Mycetophilidae clade (node J), with the groups most traditionally included (e.g. Vockeroth 1981), also included Pterogymnus as suggested by Chandler (2002), and Sciaropota, and appeared as the sister clade of the sciarid clade (node I). Jaschhof (2004) discussed the difficulty of using the highly contradictory evidence of the genitalic characters for inferring the phylogeny among the genera of the Heterotricha group. Our parsimonious solution, based only on the extant taxa, differed from both his and Chandler s (2002) conclusions but agreed with both in the obvious non-monophyly of the group. Our hypothesis on the position of the Cecidomyiidae (node H) is not in accordance with the views of Matile (1990, 1997), Blaschke-Berthold (1994), and Blagoderov & Grimaldi (2004) who placed the Cecidomyiidae as the sister-group of all the other Sciaroidea studied by them (but see the fossil analysis below). Our result also strongly contradicts the hypotheses of Wood & Borkent (1989) and Oosterbroek & Courtney (1995) who regarded the Cecidomyiidae and Sciaridae as sister-groups (a relationship that is also still shown in the Tree of Life on the Internet). In these latter phylogenies, the evidence for the Cecidomyiidae + Sciaridae relationship was based mainly on a few larval or cytological characters, which are only known in a few scattered exemplar taxa of the present ingroup. The only macromorphological character of the adults used by these authors is the costalization of the wing veins and the holoptic compound eyes but so far as our material is concerned, this cannot be regarded as a synapomorphy of the Cecidomyiidae + Sciaridae. Hennig (1973) had serious doubts about the cytological evidence for the Cecidomyiidae + Sciaridae sister-group relationship. Jaschhof & Didham (2002) regarded the Cecidomyiidae + Rangomaramidae to be the sister-group of the Sciaridae. The association of the Rangomaramidae and Cecidomyiidae by Jaschhof & Didham (2002) was based on one synapomorphy, the lack of tibial spurs, but this is not actually correct because the plesiomorphic state, the presence of spurs, is present in at least one species of Rangomarama. The sciarid clade (node K) included four main groups/clades: the Sciarosoma group/clade, the Sciarotricha group/clade, the Rangomarama group/clade and the Sciara group/clade. The position of Sciarosoma as the sister-group of all the rest of the sciarid clade was relatively weakly supported by two characters, one of which, the roundly curved anteroventral margin of laterotergite (13: 1), is unique. The support for Sciarotricha as the sister-group of the Rangomarama + Sciara group was stronger, based on four character changes, of which the occurrence of the developed pleural pit (character 16: 1) is unique (node L). The Rangomarama group was ranked as the family Rangomaramidae and postulated to be the sister-group of the Cecidomyiidae by Jaschhof & Didham (2002) (see comment under Cecidomyiidae above). According to the present phylogeny, Rangomarama is the sister-group of the Sciaridae in the strict sense. Although this relationship was not strongly supported by character evidence, it was the only parsimonious solution obtained from the present data. The second analysis, with the fossil Sciaroidea included in the ingroup, also produced two most parsimonious cladograms, the strict consensus of which (268 steps, CI 32, RI 69) was fully resolved except for two trichotomies (Fig. 2). The solution was partly identical, but partly greatly different, from the one with only the extant taxa. In the fossil cladogram, the Cecidomyiidae appeared in a different clade at the base of the cladogram, separated from all the other groups (node a). In the

12 132 Hippa, H. & Vilkamaa, P. INSECT SYST. EVOL. 36:2 (2005) other major clade, most fossil taxa were in the same apical clades together with Bolitophila, Diadocidia, Ditomyia and the Keroplatidae (node b). The fossil Mangasidae, originally described as a subfamily of the Bolitophilidae, is polyphyletic: the type genus belongs to the keroplatid clade (node c), but Mangasinae gen. sp. of Kovalev (1986) is the sister-group of Bolitophila (node d). Chandler (2002) noted that Mangas does not belong to the Bolitophilidae. The core of the Heterotricha group appeared monophyletic, provided that Sciaropota and Sciarosoma are excluded. Ohakunea + Colonomyia appeared closer to Sciaropota + Mycetophilidae (node e). The recently described Mesozoic genus Thereotricha (Blagoderov & Grimaldi 2004), assigned to Sciaroidea incertae sedis, probably belongs to the Sciaropota + Mycetophilidae clade because of the wholly setose scutum. We did not include Thereotricha in our matrix because we could not interpret many of the wing characters. The sciarid clade as a whole (node f), with the inclusion of Sciarosoma and the new Sciarotricha, was identical with the extant analysis except that the Archizelmira group now appeared as the sister-group of the Sciara group. An identical result was obtained when only the Archizelmiridae from among the fossil taxa was included in the analysis. According to the present parsimonious result, Rangomarama is the sister-group of the Sciaridae in the strict sense + the Archizelmira group (node g), based on the non-setose wing vein CuA2 (53: 1). However, this character state has a reversal into the setose type in Sciara. The Sciara group or Sciaridae in the traditional meaning is monophyletic (node i). The Mesozoic archizelmirids are monophyletic (node j), as suggested by Grimaldi et al. (2003), and they are the sister-group of the Sciara group (node h). However, the characters of the Archizelmira group are poorly known and we would not be surprised if this sister-group relationship just with the Sciara group would later be shown to be incorrect. Grimaldi et al. (2003) considered many characters common to the Sciaridae and Archizelmiridae as convergences, e.g. the wing venation, including the basal displacement of the cubital veins, and the well-developed pulvilli. We analysed the wing venation characters in detail, whereas the characters of the tarsal claws and pulvillar structures would need a SEM examination of all the groups in question and were excluded. Contrary to our results, Grimaldi et al. (2003) also considered the Archizelmiridae, together with Heterotricha and Ohakunea, for example, to be the basal groups of the Sciaroidea. However, they did not present a numerical matrix or a rigorous parsimony analysis to support their views on the position of the different sciaroid groups. Blagoderov & Grimaldi (2004) derived the wing venation of the Archizelmiridae from that of the Ditomyiidae. The results of the extant and fossil analyses taxa were in part rather different. The high level of homoplasy in the data must at least partly explain the great effect of taxon sampling on the cladogram topology. It is hard to say which of the two results is a better phylogeny. The fossil analysis is better in having more recognized taxa of the study group included, a drawback is that many of the characters were not available for study in the fossils. Interestingly, the sciarid clade (node f in Fig. 2) remained stable, the Archizelmira clade always being the sister clade of the Sciara clade, whereas other parts of the cladogram changed when inclusion or exclusion of one or more of the other sciaroid fossils were tried in analyses. However, if fossils are included in the analysis, it seems difficult to judge why representatives of all the main groups available for study should not be included. Ranking of the clades and redefinition of the Sciaridae The currently accepted family classification of the Sciaroidea seems for the most part to be cladistically well founded, so far as can be determined from the study of only a few representative taxa. It may be a matter of opinion as to whether it is necessary to separate the Ditomyiidae from the Keroplatidae at the family level. The hierarchical level of the internal classification of the mycetophilid clade remains open because only a limited variety of the extant and fossil taxa were included in the analysis. It seems that the Mesozoic Mesosciophilidae is an integral part of Mycetophilidae s. l., not a sister-group of any larger clade. Our results support Chandler s (2002) suggestion that it is an available family or subfamily assignment for the recent similar Pterogymnus. If the Mesosciophilidae is accepted as a family, then the family level status for the Lygistorrhinidae is also correct, although we suggest that there is a closer relationship between the Mycetophilidae and Lygistorrhinidae, as Blagoderov & Grimaldi (2004)